Northland’s job ads are up. Its employment is flat. Its unemployment rate went from 5.4 to 8.8 percent.
Four official series describe work in Northland over the year to June 2026, and they do not tell one story. Job advertisements rose. Filled jobs fell. The number of people employed barely moved. And the unemployment rate went from 5.4 to 8.8 percent, the largest increase of any region in the country. This article lays the numbers side by side, rules out the easiest explanation, and is honest about what it cannot tell you.
What we are claiming, and what we are not
The claim this article makes is narrow: four independent official measures of work in Northland moved in directions that do not fit together, and the most obvious explanation does not hold. That is an observation about published statistics, each of which you can check against its source below.
Here is what we are not claiming. We do not know why this is happening, and we do not pretend to. We are not evaluating any policy, any government, or any council; 2026 is an election year, and these figures will be argued over by people with positions to defend. Ours is stated at the end, and it is commercial, not political. If you quote this article, quote the numbers with their dates and sources, because that is all it establishes.
Four numbers that do not agree
All four cover the year to June 2026, from three separate government sources, measured in three different ways: a household survey, payroll data from IRD, an index of online job ads, and benefit records. That independence matters. A single series can mislead; four series with different methods, all pointing at the same region, are describing something real even when they disagree about what.
- Unemployment rate: 5.4% to 8.8% (Stats NZ, Household Labour Force Survey, Table 6, not seasonally adjusted, annual comparison as Stats NZ itself recommends for regional data). The largest increase of any region in New Zealand, and it is not close: the next largest was Gisborne/Hawke’s Bay at +2.1 percentage points.
- Job advertisements: up 4.0% (MBIE Jobs Online, an index across Seek, Trade Me Jobs, Education Gazette and Kiwi Health Jobs). Advertised demand for workers in Northland grew over the same year.
- Filled jobs: down 2.2%, about 1,600 jobs (Stats NZ Monthly Employment Indicators, derived from IRD payday filing, so this is administrative data rather than a survey). Northland was the only major region where filled jobs fell while Canterbury, Auckland and Waikato all grew.
- Jobseeker Support: 11.9% of the working-age population, the highest rate in the country against a national 6.8%, and the largest annual increase (MSD Benefit Fact Sheets, June 2026 quarter).
Advertised work rising while filled work falls and benefit receipt climbs is not what a simple “there is no work” story looks like, and not what a simple “there is plenty of work” story looks like either. Something between the advertisement and the filled job is not connecting.
Every region, same measure, same year
Context is what stops a single striking number from misleading. Here is the same unemployment measure for every region, June 2025 to June 2026. Several regions improved. Canterbury’s rate fell by 1.3 percentage points. Northland is not an extreme case of a national trend; it is an outlier even against a soft year.
Stats NZ, HLFS Table 6, not seasonally adjusted; annual comparison per the table’s own guidance. Regional survey estimates carry sampling error, so treat small movements (a few tenths) as noise; Northland’s +3.4 points is far outside it.
The clearest split in the country runs between the islands. North Island unemployment was 6.0% in June 2026, up from 5.3% a year earlier; the South Island was 3.7%. So there is a genuine North Island pattern, and Northland is its sharpest case rather than its explanation.
It is not a jobs-lost story
The detail that reframes the headline: Northland’s employed count barely moved in net terms. 90,000 people in June 2025, 90,100 in June 2026. What changed is that the labour force grew by roughly 3,500 people while net employment grew by about 100, and nearly all of that growth shows up as unemployment.
That is a description of the arithmetic, not a diagnosis, and it has a limit worth stating: these are net totals. A flat employed count is consistent with a stable workforce, and equally consistent with substantial job losses offset by hiring elsewhere; survey stock figures cannot tell those apart. What the arithmetic does establish is that the rise in the rate is accounted for by labour-force growth rather than by a net fall in employment. It does not tell you who the additional searchers are or why they entered the search now, and neither does any series we have.
The numbers above, read against advertising
Read on their own, the unemployment and underutilisation rates above could suggest work simply is not there. The advertising data complicates that, and has for a decade:
One caution in both directions: every region’s advertising is still well below three years ago (Northland is down about 31% on 2023; Auckland about 43%). The national index grew 7% in the year to June 2026, the fourth consecutive quarter of annual growth, after eleven consecutive quarters of decline. “Recovering from a deep trough” is supportable. “Strong” is not. And an ad is not a vacancy: the index counts advertisements, so a hard-to-fill role re-advertised repeatedly registers growth without any new job existing. That caveat matters for exactly the pattern this article is about, and it is one of the untested explanations below.
The easiest explanation, ruled out
The tidy way to reconcile falling filled jobs with rising ads would be people leaving: fewer working-age residents means fewer filled jobs without anyone losing work. So we checked, against Stats NZ’s subnational population estimates.
Northland’s working-age population (15 to 64) was 117,200 in June 2024 and 117,200 in June 2025. Flat to the hundred. Total population grew by 760 people, net internal migration was positive (more New Zealanders moved in than out), and all of the net growth was in the 65-plus age band, while the under-15 count fell. This is a region ageing in place, not emptying out. “Working-age people left” is not supported.
The honest limitation: the subnational estimates run a year behind the labour data, so this rules out population decline in the year leading into the divergence, not during it. The June 2026 estimates publish around late October, and we will check them when they do. One movement inside those figures is worth naming without interpreting: net international migration into Northland fell from 1,600 to 70 in a single year. We do not know whether that connects to anything above, and testing it would need data nobody publishes at this level.
What could explain it, none tested
A gap between advertised work and filled work is consistent with several things, and we have not tested any of them: a mismatch between the skills advertised for and the skills held; distance and transport across a large, low-density region; advertised pay against housing and living costs; employers re-advertising roles they cannot fill, which inflates an ad index without any new job existing; and measurement artefacts, since the filled-jobs data is provisional for three months and an ad is not a vacancy. Each of these is a hypothesis. Choosing one without evidence would be exactly the kind of claim this article exists not to make.
What would actually settle it: the regional series over several years rather than one; Northland’s industry mix against the national pattern; the October population estimates; and filled jobs broken down by who is filling them, which we did not find in any Stats NZ, MBIE or MSD release checked for this article. If we do that work, we will publish what it shows, whichever story it supports.
Who the national rise landed on
Nationally, unemployment rose from 5.3% to 5.6% over the quarter and underutilisation from 12.9% to 13.8%. That rise was not spread evenly, and the sharpest movement in the entire release is by age:
One figure moved the other way, and the pairing is instructive: the youth NEET rate (15 to 24 year olds not in employment, education, or training) improved to 13.8% from 14.4% the quarter before. That is not a contradiction. NEET counts people outside both work and study; unemployment counts people outside work who are actively looking. Both numbers can be honestly quoted to tell opposite stories about young people, which is a small demonstration of this article’s larger point: the label on the number decides the story, so check the label.
The burden is uneven by ethnicity too. Stats NZ reports the Māori underutilisation rate at 22.8%, up from 19.5% a year earlier; Pacific peoples at 22.1%, against 12.5% for Asian and 11.8% for European New Zealanders (not seasonally adjusted). And it is uneven against prices: the same release sets wage growth of 2.0% for the year beside CPI inflation of 4.1%. We quote both numbers as published and leave the arithmetic to the reader.
The churn the headline number hides
The Jobseeker Support total moved by just 2,472 over a year, which sounds like a still pond. The flow underneath it is anything but:
Forty-three thousand people were granted Jobseeker Support in a single quarter, against a total stock of 218,481; tens of thousands left over the same weeks, into work and otherwise. The exits and cancellations figures cover all main benefits, not Jobseeker alone, so the columns do not reconcile into a neat ledger, and we are not building one. The point survives the caveat: a nearly-flat total is not a nearly-still population. People move through these numbers constantly, in both directions, and each entry is someone starting a job search under the conditions the rest of this article describes.
The same trap, nationally: three numbers that all describe “people who need work”
Northland is the sharpest case of something that trips up national coverage constantly: the official numbers that sound interchangeable are not. For the same June 2026 quarter, all from official sources:
The relationships between them are not what the labels suggest. The unemployed count is one component of the underutilised total, by definition. But neither of the first two contains the other: someone with a working partner can be unemployed and ineligible for the benefit, and someone on Jobseeker Support with a health condition may not be counted as unemployed. Quoting the 218,481 as “jobseekers” without the Work Ready split folds in roughly 98,000 people MSD does not class as Work Ready; all of that total’s net growth over the year was in the health condition and disability category, while the Work Ready group shrank. And the 440,000, the broadest of the three measures, rose 9.5% in a year and now stands at two and a half times the unemployment number.
This is also why “the labour market is recovering” and “it is getting harder for people looking for work” are both currently true. Job ads have grown for four straight quarters. Unemployment and underutilisation rose anyway. Which one you hear depends on which number the speaker picked.
Why we are publishing this
Rolebird sells job-application preparation, so an article observing that advertised jobs and filled jobs are not connecting sits close to our commercial interest, and you should read it knowing that. We have tried to earn trust the only way that works: every figure above is from the issuing agency’s own release, dated, and checkable without taking our word for anything. We claim no cause, and we do not claim that any product, ours included, changes these numbers for the people in them.
If we got a figure wrong, we will correct it and say what changed; minor wording fixes get a dated note. That is the same corrections practice as the rest of our research.
Sources
All accessed 5 and 6 August 2026. Regional HLFS figures are not seasonally adjusted; Stats NZ recommends annual comparisons for them, which is what this article uses. On 5 August 2026 Stats NZ also published corrections to two series in earlier quarters (an HLFS hours-reason series and one public-sector LCI series); neither affects any figure quoted here.
- Stats NZ, Labour market statistics: June 2026 quarter (released 5 August 2026), including Table 6, key labour market measures by regional council area, and the release’s own LCI/CPI comparison.
- Stats NZ, Employment indicators: June 2026 (released 28 July 2026), filled jobs from payday filing. Actual data provisional for three months.
- Stats NZ, Subnational population estimates: At 30 June 2025 (provisional) (released 29 October 2025), Tables 1, 3 and 5.
- MBIE, Jobs Online, June 2026 quarter: All Vacancies Index and the regional growth series (annual, three-year and ten-year changes). MBIE notes the quarterly series is no longer seasonally adjusted and recommends annual comparison.
- MSD, Benefit Fact Sheets Snapshot: June 2026 quarter: Jobseeker Support national and regional rates and composition, grants, exits into work, and main benefit cancellations.